Introduction
Nasu-Hakola disease (NHD), also referred to as polycystic lipomembranous osteodysplasia with sclerosing leukoencephalopathy (PLOSL; OMIM 221770, OMIM 605086), is a rare autosomal recessive disorder characterized by a progressive presenile dementia and bone cysts. Approximately 200 NHD cases have been reported worldwide, primarily in Japan and Finland, with few cases also reported in Italy [
2,
7,
39,
44,
69,
82,
96].
Clinically, patients with NHD show recurrent pathological bone fractures, with frontal lobe syndrome and progressive dementia during the fourth decade of life [
63,
77]. Pathologically, the brains of NHD patients exhibit extensive demyelination, astrogliosis, axonal loss, accumulation of axonal spheroids, calcification, and activation of microglia in the white matter of frontal and temporal lobes and the basal ganglia [
3,
77]. Cortical deposition of amyloid beta (Aβ) and focal neocortical neurofibrillary pathology have been also described [
33]. NHD is caused by homozygous pathogenic mutations in the genes
triggering receptor expressed on myeloid cells 2 (
TREM2) or
TYRO protein tyrosine kinase binding protein (
TYROBP), alternatively named
DNAX-activation protein 12 (
DAP12) [
78,
79,
83]. Seventeen homozygous pathogenic mutations have been identified in
TREM2 or
DAP12 genes and are predicted to cause loss of function and lead to a similar disease phenotype [
50,
52,
57,
105]. More recently, missense heterozygous variants in
TREM2 have been implicated in risk for late-onset Alzheimer’s disease (AD) [
37,
42] and frontotemporal dementia [
9,
38]. Thus, understanding the functional impact of
TREM2 gene variants has major implications for a number of dementias.
The TREM2 receptor is expressed in myeloid cell populations, including peripheral macrophages and microglia in the central nervous system (CNS) [
47,
91]. TREM2 is a type I trans-membrane glycoprotein with an extracellular V-type immunoglobulin (Ig) ectodomain, a connecting stalk followed by a transmembrane region and a C-terminal tail [
51]. At the cell surface, TREM2 receptor signaling occurs through association with the adaptor protein DAP12, which interacts with the transmembrane domain of TREM2. The TREM2 receptor has been shown to bind bacterial components [
19], anionic and zwitterionic lipids [
102], and myelin [
88]; yet, the endogenous ligand remains poorly defined. Activation of the TREM2/DAP12 complex mediates the recruitment of the protein tyrosine kinase Syk, resulting in the phosphorylation of downstream mediators such as PLC-γ, PI3K and Vav2/3, which lead to mTOR and mitogen-activated protein kinase (MAPK)-mediated downstream signaling [
16,
85,
101]. These intracellular signals promote macrophage/microglia survival [
75,
76,
102]; proliferation [
76]; phagocytosis [
97]; synapse elimination [
28]; cellular metabolism [
101]; and autophagy [
101]. Proteolytic cleavage of the TREM2 ectodomain and/or translation of an alternatively spliced
TREM2 transcript [
20,
27,
41,
68,
95,
103] are responsible for the production of a soluble form of the receptor (sTREM2) which is detected in the cerebrospinal fluid (CSF) [
27,
86]. Some of the NHD-causing missense mutations in
TREM2 likely affect TREM2 protein folding and stability [
53], but the cellular mechanisms and the pathways dysregulated during the disease are still unknown. Despite the predicted loss of function effect of NHD mutations in
TREM2, mouse models of
Trem2 deficiency fail to capture key pathological aspects of NHD. As such, there remains a critical gap in our understanding of the pathologic impact of NHD mutations in the CNS and our ability to develop novel therapeutics.
Human stem cell models have emerged as a powerful cellular system that enables modeling of rare mutations in the cell-types affected in disease, including those that cause NHD [
12,
18,
32,
73]. Yet, the extent to which stem cell models capture disease relevant phenotypes remains uncertain. To begin to define the impact of a rare loss of function mutation in
TREM2 on microglia function, we generated macrophages and human induced pluripotent stem cell (iPSC)-derived microglialike cells (iMGLs) from two families affected by NHD. The three affected individuals are homozygous for the
TREM2 p
.Q33X mutation, two unaffected heterozygous
TREM2 p.Q33X parents, a healthy sibling that is homozygous for the wild-type allele (
TREM2 WT), and two unrelated controls (
TREM2 WT).
TREM2 p.Q33X introduces an early stop codon and is subject to nonsense mediated decay; hence, there is no TREM2 protein product [
80]. Herein, transcriptional profiling and functional analyses of the
TREM2 p.Q33X mutation reveal that it leads to dysregulation of lysosomal function, lipid metabolism and microglia activation, and these phenotypes are recapitulated in brain tissues from NHD patients. Finally, we find that targeting lysosomal dysfunction in an mTOR-dependent or independent manner rescues lysosomal, lipid and activation state defects caused by the
TREM2 mutation.
Materials and methods
Patient consent
Skin biopsies were collected following written informed consent from the control unaffected sibling or the parents for the NHD affected individuals. The informed consent was approved by the Washington University School of Medicine Institutional Review Board and Ethics Committee (IRB 201104178).
Human postmortem brain tissue
Human tissue sections were obtained from two NHD patients, three control patients and two MS patients. The two MS patients were obtained from The Neuroinflammatory Disease Tissue Repository at Washington University St. Louis. Brain autopsies from NHD tissues were obtained from the University of Washington Neuropathology Core Brain Bank and the Meiji Pharmaceutical University, Tokyo, Japan. Under protocols approved by the Institutional Review Boards of the University of Washington and the Meiji Pharmaceutical University, all patients had previously given informed consent to share and study autopsy material. All methods for processing and analyzing the brain autopsy tissues followed relevant guidelines and regulations. Demographic and clinical characteristics of the donors of human brain tissues at the time of collection are indicated in Supplementary Table 1, online resource.
Dermal fibroblast isolation
Dermal fibroblasts were isolated from skin biopsies obtained from research participants from two NHD families. Briefly, skin biopsies were collected by surgical punch and stored in Fibroblast Growth Media (Lonza). To isolate dermal fibroblasts from skin biopsy, the biopsies were rinsed with PBS and cut lengthwise with dissecting scissors. The resulting tissue sections were then plated into a dry 24-well tissue culture treated plate (approximately 6–12 sections). After removing excess PBS from the wells, 300µL of fibroblast growth media (Lonza) was carefully added and tissue was incubated at 37°C and 5% CO2. After 24 h, tissue was supplemented with 1 mL fibroblast growth media and media changes were repeated every 3–4 days. Fibroblast cells were observed to migrate from the tissue within 2 weeks of culture. Dermal fibroblasts were maintained in Fibroblast Growth Media (Lonza) supplemented with penicillin/streptomycin.
Macrophage differentiation from PBMCs
Peripheral blood mononuclear cells (PBMCs) (IDE: SB CTRL, IDO: NHD1, and 1F1: NHD2) were purified from human blood on Ficoll-Paque PLUS density gradient (Amersham Biosciences, Piscataway, NJ). To generate macrophages, PBMCs were cultured in 6-well culture plates (3 × 106 cell/well) in RPMI-1640 without fetal bovine serum. After 2 h of culture, PBMCs were washed twice with PBS 1X and cultured in RPMI supplemented with 50 ng/mL MCSF for 7 days at 37°C with 5% CO2.
iPSC generation and characterization
Human fibroblasts (IDE: SB CTRL, F12455: NR CTRL, F14532: NR CTRL2, IDO: NHD1, and 1F1: NHD2, F21675: NHD3, F21673: HET1, F21674: HET2) were transduced with non-integrating Sendai virus carrying the four factors required for reprogramming into iPSC: OCT3/4, SOX2, KLF4, and cMYC [
6,
98]. Single colonies showing morphological evidence of reprogramming were isolated by manual dissection. Human iPSCs were cultured using feeder-free conditions (Matrigel, BD Biosciences, Franklin Lakes, NJ, USA). Human iPSCs were thawed (1–2 × 10
6 cells/mL), diluted in DMEM/F12, and centrifuged at 750 rpm for 3 min. The resulting iPSC pellet was then diluted in mTeSR1 supplemented with Rock inhibitor (Y-27632; 10 µM final). iPSCs were subsequently cultured in 37°C, 6% CO
2 with daily medium changes (mTesR1, STEMCELL Technologies, Vancouver, BC, CA). The cell lines were regularly tested for mycoplasma. All iPSC lines were characterized using standard methods [
98]. Each line was analyzed for chromosomal abnormalities by karyotyping (Supplementary Fig. 1a, b, online resource), for pluripotency markers (OCT4A, SOX2, SSEA4, TRA1) by immunocytochemistry (ICC) (Invitrogen A24881) (Supplementary Fig. 1c, d, online resource), and for
TREM2 mutation status (homozygous and heterozygous for the
TREM2 p.Q33X mutation) by Sanger sequencing (Supplementary Fig. 1e, f, online resource).
iPSC-derived microglia-like cells (iMGLs)
iMGLs were generated as previously described in [
1,
25,
66]. iPSCs were differentiated into hematopoietic precursors cells (HPCs) using a STEMdiff Hematopoietic kit (STEMCELL Technologies) and following manufacturer’s instructions. Briefly, to begin HPC differentiation, iPSCs were detached with ReLeSR™ (STEMCELL Technologies) and passaged in mTeSR1 supplemented with Rock inhibitor to achieve a density of 100–200 aggregates/well (for iPSCs from mutation carrying donors, aggregate numbers to be plated requires line-specific optimization). On day 0, cells were transferred to Medium A from the STEMdiff Hematopoietic Kit. On day 3, flattened endothelial cell colonies were exposed to Medium B and cells remained in Medium B for 10 additional days while HPCs began to lift off the colonies. After 12 days in culture, HPCs were collected by removing the floating population with a serological pipette and the adherent portion after incubation with Accutase™ (STEMCELL Technologies) for 15 min at 37°C. The floating and adherent CD43
+ population (see below
Fluorescent Activated Cell Sorting (FACS) section) was sorted with a Becton Dickinson FACSAria II cell sorter. At this point, HPCs can be frozen in CryoStor® CS10 (STEMCELL Technologies) and stored in liquid nitrogen. iMGL induction was achieved by culturing CD43
+ HPCs in iPSC-Microglia medium (DMEM/F12, 2X insulin-transferrin-selenite, 2X B27, 0.5X N2, 1X glutamax, 1X non-essential amino acids, 400 mM monothioglycerol, and 5 mg/mL human insulin) freshly supplemented with 100 ng/mL IL-34, 50 ng/mL TGFβ1, and 25 ng/mL M-CSF (Peprotech) for 25 days (37 days from iPSC). During the last 3 days in culture, 100 ng/mL CD200 (Novoprotein) and 100 ng/mL CX3CL1 (Peprotech) were added to iPSC-Microglia medium to mimic a brain-like environment (Supplementary Fig. 2A, online resource). All the subsequent analyses described in the paper have been performed on fully mature iMGL (between day 40 to day 52) unless otherwise stated. All the analyses performed in this work were run on the first NHD family (IDE: SB CTRL, F12455: NR CTRL, IDO: NHD1, and 1F1: NHD2) in parallel, and on the second NHD family (F14532: NR CTRL2, F21675: NHD3, F21673: HET1, F21674: HET2) in parallel, except for RNAseq, electron microscopy and LPS treatment, which were run on IDE: SB CTRL, IDO: NHD1, and 1F1: NHD2.
Fluorescent activated cell sorting (FACS)
All steps were performed on ice or using a pre-chilled refrigerated centrifuge set to 4˚C with all buffers/solutions pre-chilled before addition to samples. For HPC sorting, HPCs were collected using sterile filtered FACS buffer (1X DPBS, 2% BSA), cells were then filtered through 70 μm filters to remove large clumps, washed with FACS buffer (300 × g for 5 min 18 C), then stained for 20 min at 4˚C in the dark using the following antibodies: CD43-APC (Cat#: 343206, Clone: 10G7), CD34-FITC (Cat#: 343504, Clone: 581), CD45-Alexa 700 (Cat#: 304024; Clone: H130) from BioLegend, and Zombie Aqua™ Fixable Viability Kit (BioLegend). The CD43+ total population was sorted with a Becton Dickinson FACSAria II cell sorter. For detection of microglial surface markers, iMGLs were incubated for 10 min on ice with anti-CD16/CD32 to block Fc receptors (1:50; Miltenyi Biotec, Cat #:120-000-442) and with Zombie Aqua™ to identify live cells. Then, iMGLs were stained with CD11b-PeCy7 (Cat#: 101216; Clone: M1/70), CD45-Alexa 700 (Cat#: 304024; Clone: H130), CD80-BV421 (Cat#: 305222, Clone: 2D10), CD86-PerCP/Cy5.5 (Cat#: 30420, Clone: IT2.2), MERTK (Cat#: 367620, clone: 590H11G1E3) all from BioLegend, CD14-PeCy7 (Invitrogen, Cat#: 25-0149-41, Clone: 61D3), HLA-DR-PECF594 (BD, Cat#: 562331 clone: G46-6), TREM2-APC (R&D, FAB17291A), TREM2-biotinilated (Clone: E29E3, generously provided by Dr. Marco Colonna) for 20 min at 4˚C in the dark. AnnixinV/Propidium iodide positive cells were detected with a FITC Annexin V Apoptosis Detection Kit with PI (Biolegend). Cells were acquired on a BD LSRFortessa and BDX20 and data analyzed with FlowJo software (FlowJo).
Lysosomal acidity measurement
To evaluate acidic vesicles, iMGLs were incubated with 5 nM of LysoTracker® Red DND-99 (ThermoFisher, L7528), diluted in the cell medium at 37°C for 5 min. Live cell images of NR CTRL, SB CTRL, NHD1 and NHD2 were acquired with a Nikon Eclipse 80i fluorescent microscope and Metamorph Molecular Devices software. Live cell images of NR CTRL2, HET1, HET2 and NHD3 were acquired with Zeiss LSM980 Airyscan 2 laser scanning confocal microscope (Carl Zeiss Inc., Thornwood, NY) at 40x. A microtubule probe (ViaFluor® 488, Biotium, 70062-T) was used to define cellular structure. Fluorescence intensity of LysoTracker Red was quantified by Fiji. Briefly, the fluorescence intensity of LysoTracker Red in the soma of each cell was measured and then corrected for background fluorescence resulting in the Corrected Total Cell Fluorescence (CTCF) values. To study the overall protease activity, iMGLs were incubated for 4 h at 37°C with 10 µg/mL of DQ™ Red BSA (ThermoFisher, D12051) diluted in the cell medium. Cells were washed once with HBSS solution. Live and fixed cell images were acquired with the laser confocal microscope ZEISS LSM 980 with Airyscan. Fluorescence intensity of DQ™ Red BSA staining in the soma was quantified by Fiji as previously described (Marwaha and Sharma, 2017). DQ™ Red BSA stained cells were next fixed for further immunofluorescence analyses.
Myelin production and iMGL treatment
Human myelin was prepared as previously described [
15,
71] and stored in lyophilized form at – 80°C. Prior to use, myelin was suspended in DMEM to a final concentration of 2 mg/mL and dissolved by vortexing and sonicating. Myelin was then irradiated with 10,000 RADS to achieve sterility. Aliquots were stored at – 80°C for further use. For treatment with myelin, iMGLs were plated on Matrigel-coated 12-mm coverslips at 5 × 10
4 cell/well and incubated with 10 μg/mL of pHrodo-labeled myelin (pHrodo™ Red, SE, ThermoFisher) for 24 h.
Torin1 and curcumin analog C1 treatment in iMGLs
Fully mature iMGLs were plated in 24-well plate at a density of 1 × 105 cell/well and treated with 250 nM Torin 1 (Tocris, 4247) for 4 h or with 1 μM of curcumin analog C1 (Cayman Chemical Company, 34255) for 7 h or with DMSO as untreated control. After 4 h and 7 h of treatment respectively, iMGLs were washed once in PBS and cultured in complete iPSC-Microglia medium for 24 h. Cells were then collected and stained for detection of microglial surface markers by FACS or for PLIN2 by immunocytochemistry (ICC).
Immunocytochemistry
Cells were washed three times with DPBS (1X) and fixed with cold PFA (4% w/v) for 20 min at room temperature (RT) followed by three washes with PBS (1X). Cells were blocked with blocking solution (PBS with 0.1% Triton X-100 and 1% BSA) for 30 min at RT. Primary antibodies were added at respective dilutions (see below) in blocking solution and placed at 4°C overnight. The next day, cells were washed 3 times with PBS for 5 min and stained with Alexa Fluor conjugated secondary antibodies from Invitrogen (1:800) for 2.5 h at room temperature in the dark. After secondary staining, cells were washed 3 times with PBS and cover slipped with ProLong™ Diamond Antifade Mountant or Fluoromount-G™ (ThermoFisher). Primary antibodies used for ICC: anti-P2ry12 (1:500, HPA014518 Sigma), anti-TREM2 (1:200, AF1828 R&D Systems), anti-CD68 (1:100, M0718 Dako), anti-TMEM119 (1:100, ab185333 Abcam), anti-IBA1 (1:500; Wako, 019-19741).
Lipid droplets staining
For Perilipin 2 (PLIN2) staining, cells were blocked in PBS with 0.1% saponin and 1% BSA for 5 min and subsequently incubated with ADRP/Perilipin 2 antibody (1:500, 15294-1-AP Proteintech) in PBS with 0.1% saponin and 1% BSA overnight at 4°C. For BODIPY staining, cells were incubated in PBS with BODIPY FL Dye (1:1000 from a 1 mg/mL stock solution in DMSO; ThermoFisher) a dye that specifically labels neutral lipids [
89] and Hoechst 33342 (1:5000; ThermoFisher) for 20 min at room temperature (RT). To analyze the percentage of lipid-droplet-containing iMGLs, the numbers of total Hoechst + cells and of Hoechst + cells with BODIPY + lipid droplets were counted, and the percentage of BODIPY + iMGLs was calculated. For confocal analysis, images were acquired with a Zeiss LSM980 Airyscan 2 laser scanning confocal microscope (Carl Zeiss Inc., Thornwood, NY) equipped with 63X and 40X, 1.4 numerical aperture (NA) Zeiss Plan Apochromat oil objectives. The system was equipped with a unique scan head, incorporating a high-resolution galvo scanner along with two PMTs and a 32-element spectral detector. ZEN 3.4.9 Blue edition software was used to obtain Z-stacks through the entire height of the cells. Images taken were optimized for 1 airy unit using the 405 nm diode, 488 nm diode, and 561 nm diode and 633 nm HeNe (helium neon) lasers. Images were finally processed with ImageJ and Imaris Software (Bitplane, Switzerland). Quantification was performed on original orthogonal Z-projections generated in ImageJ software. The particle numbers were quantified with ImageJ 1.5j8 (NIH) with size (pixel2) settings from 0.1 to 10 and circularity from 0 to 1. PLIN2-positive droplets with intensity above cytoplasmic background and size (pixel2) from 10 were quantified. A total of 40–50 cells per line were analyzed.
Histology and immunohistochemistry of formalin-fixed human brain tissue
Formalin-fixed paraffin embedded (FFPE) 5 μm human tissue sections were stained with solochrome cyanine to detect area of demyelination. For immunohistochemistry analysis, sections were deparaffinized and rehydrated in xylene, then sequential concentrations of ethanol to water followed by antigen retrieval in boiling 0.01 M citric acid pH 6.0 for 10 min. Sections were then incubated in PBS with 5% horse serum and 0.1% triton-X for 1 h at RT followed by incubation overnight at 4°C with primary antibodies: anti-Iba1 (1:250; Novus, NB100-1028), anti-TMEM119 (1:100, ab185333 Abcam), anti-PLIN2 (1:100; 15294-1-AP, Proteintech), anti-CD68 (1:50, M0718 Dako), and monoclonal anti-LAMP1 (H4A3, undiluted, developed by Developmental Studies Hybridoma Bank). After primary antibody incubation, sections were washed three times in PBS and incubated in PBS with Alexa Fluor conjugated secondary antibodies from Invitrogen (1:1000) for 1 h at RT in the dark. Sections were mounted with Vectashield (Vector Laboratories, H-1000). Quantitative evaluation of microglial cell morphology in tissue sections was performed as described in the literature using ramification index (RI) calculated by the following equation: 4π × cell area/ (cell perimeter)
2 [
13]. The RI of perfectly round cells is 1; if morphology deviates from circular form, RI is smaller than 1; when the cell is highly ramified the RI is close to zero. The images were acquired with a Nikon Eclipse Ni fluorescent and bright field microscope equipped with 10X, 20X, and 60X zoom objectives. LAMP1 and PLIN2 were analyzed based on the percent area of LAMP1 + Iba1 + and PLIN2 + Iba1 + staining (number of positive pixels/mm2) and then normalized to the percentage of Iba1 + (number of positive pixels/mm2) within the region of interest. CD68 was analyzed as the percentage of CD68 + TMEM119 + Iba1 + and then normalized based on the percentage of TMEM119 + Iba1 + staining (with the assistance of NIS-Elements software).
Immunoblotting
iMGLs were lysed in RIPA buffer (50mMTris, 150 mM NaCl, 1% SDS, and 1% Triton X-100) containing PMSF, leupeptin, activated sodium orthovanidate, apoprotinin, and phosphatase inhibitor cocktail 3 (Sigma Aldrich Cat. Number P0044). Lysates were mixed with 4 × Laemmli sample buffer (Bio-Rad, Cat#: 161-0747) and 10% β-mercaptoethanol and heated at 95 °C for 10 min and run on a 4–12% bis–tris gel (Nupage). Proteins were transferred to PVDF membrane and blocked for 1 h at RT in 5% milk in phosphate buffered saline with 0.1% Tween 20 (PBS-T). Membranes were probed with the mouse anti-TREM-2 mAbs supernatants (clones 10B11 and 21E10, undiluted) and GAPDH (1:500, Thermo Fisher Scientific, Cat# MA5-15738, RRID: AB_10977387) overnight at 4°C. Membranes were subsequently washed and incubated in affiniPure Goat anti-mouse HRP (1:2000, Jackson Immuno Research Labs, Cat# 115-035-174, RRID: AB_2338512) for 1 h at RT, washed, and developed using Lumigen ECL ultra-reagent (TMA-100).
sTREM2 ELISA
A sandwich ELISA for human soluble TREM2 (sTREM2) was developed as previously described [
21]. Briefly, an anti-human TREM2 monoclonal antibody (R&D Systems, catalog no. MAB1828, clone 263602; 0.5 mg/ml) was used as a capture antibody and coated overnight at 4°C on MaxiSorp 96-well plates (Nalgene Nunc International, Rochester, NY) in sodium bicarbonate coating buffer (0.015 M Na2CO3 and 0.035 M NaHCO3, pH 9.6). Washes between the steps were done four times with PBS/0.05% Tween 20 (Sigma-Aldrich). Wells were then blocked for 4 h at 37°C with 10% fetal bovine serum (FBS) in phosphate-buffered saline (PBS). Freshly thawed supernatants and recombinant human TREM2 standard (SinoBiological, catalog no. 11084-H08H-50) were incubated in duplicate overnight at 4°C. For detection, a goat anti-human TREM2 biotinylated polyclonal antibody (R&D Systems, catalog # BAF1828; 0.2 mg/mL) was diluted in assay buffer (PBS/10% FBS at 1:3000) and incubated for 1.25 h at RT on an orbital shaker. After washing, wells were incubated with horseradish peroxidase–labeled streptavidin (BD Biosciences, San Jose, CA; diluted 1:3000) for 1 h at RT with orbital shaking. Horseradish peroxidase visualization was performed with 3,3′,5,5′ tetramethylbenzidine (Sigma-Aldrich, St. Louis, MO) added to each well for 10 min at RT in the dark. Color development was stopped by adding an equal volume of 2.5 N H2SO4. Optical density of each well was determined at 450 nm. Samples were run in duplicate in each assay. Raw values are provided as [pg/ml].
Total RNA was isolated from iMGLs cells using the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. For iMGLs cell pellets frozen in Qiazol, 200µL of chloroform was added, then the samples were shaken for 5 s and centrifuged at 12,000 g for 15 min or until complete phase separation. 400-600µL aqueous phase was then transferred into a new tube and cleaned using reagents and spin columns from the RNeasy Mini Kit. Cell line samples were DNase-treated using reagents from the RNase-Free DNase Set (Qiagen, Hilden, Germany). TapeStation 4200 System (Agilent Technologies) was used to perform quality control of the RNA concentration, purity, and degradation based on the estimated RNA integrity Number (RIN), and DV200. Extracted RNA (10 µg) was converted to cDNA by PCR using the High-Capacity cDNA Reverse Transcriptase kit (Life Technologies). Samples were sequenced by an Illumina HiSeq 4000 Systems Technology with a read length of 1 × 150 bp, and an average library size (mapped reads) of 36.5 ± 12.2 million reads per sample.
Identity-by-Descent (IDB) [
11] and FastQC analyses were performed to confirm sample identity. STAR (v.2.6.0) [
22] was used to align the RNA sequences to the human reference genome: GRCh38.p13 (hg38). The quality of RNA reads including the percentage of mapped reads and length of sequence fragments was performed. Aligned sequences were processed using SAMtools (v.1.9) [
59]. The average percentage of unique mapped reads in the BAM files was 82.5% ± 3.62, and the average percentage of mapped reads to GRCh38 was 89.77% ± 5.12. Salmon (v. 0.11.3) [
81] was used to quantify the expression of the genes annotated. Protein coding genes were selected for further analyses (Supplementary Table 2, online resource).
Principal component analyses and differential expression analyses
Human macrophages and iMGLs were plotted using the plotly package [
60] based on regularized-logarithm transformation (rlog) counts and utilizing the 500 most variable protein coding genes. Differential gene expression between WT and NHD carriers was calculated using the DESeq2 (v.1.22.2) package [
62]. The transcripts per million (TPM) of the gene set of interest were made into heatmaps with ComplexHeatmap [
36]. In iMGLs, the TPM was further normalized by the expression of
GAPDH. The top 500 genes were made into a PCA graph to investigate the differences between samples. The DEG analyses compare the expression between WT and NHD carriers in different cell types and were further made into a volcano plot with ggplot2. The significant DEGs (FDR < 0.05, unless otherwise stated) were split into upregulated and downregulated DEG, and separately searched for pathway enrichment by EnrichR [
14,
56,
104].
RNA expression analysis by NanoString nCounter
RNA was isolated from FFPE sections, as previously described [
54]. Briefly, for each sample, four 10 micron sections were processed using the Recover All RNA isolation kit (Ambion, ThermoFisher). RNA yield and fragment-length distribution were measured by Qubit Fluorimeter (Molecular Probes, ThermoFisher) and 2100 Bioanalyzer (Agilent Technologies). 1 μg of total RNA was hybridized with the PanCancer Immune panel (NanoString Technologies). Counts were normalized and log2 transformed using the nSolver 3.0 Analysis Software (NanoString Technologies).
Transmission electron microscopy
For ultrastructural analyses, iMGLs were fixed in 2% paraformaldehyde/2.5% glutaraldehyde (Ted Pella Inc., Redding, CA) in 100 mM sodium cacodylate buffer, pH 7.2 for 2 h at RT. Samples were washed in sodium cacodylate buffer and postfixed in 2% osmium tetroxide (Ted Pella Inc) for 1 h at RT. After rinsing extensively in dH2O, samples were then bloc stained with 1% aqueous uranyl acetate (Electron Microscopy Sciences, Hatfield, PA). Samples were washed in dH2O, dehydrated in a graded series of ethanol, and embedded in Eponate 12 resin (Ted Pella Inc). Sections of 95 nm were cut with a Leica Ultracut UCT ultramicrotome (Leica Microsystems Inc., Bannockburn, IL), stained with uranyl acetate and lead citrate, and viewed on a JEOL 1200 EX transmission electron microscope (JEOL USA Inc., Peabody, MA) equipped with an AMT 8 megapixel digital camera and AMT Image Capture Engine V602 software (Advanced Microscopy Techniques, Woburn, MA). Multivesicular bodies (MVB) and lipid droplets analyses were performed with ImageJ (Fiji) and the number of MVB and lipid droplets was counted per cell area.
Discussion
Here, by combining transcriptomic and functional analyses in control and NHD iMGLs and brain tissues from NHD patients, we further improved our understanding of the mechanisms underlying this rare leukodystrophy. We showed that iMGLs derived from patients affected by NHD and carrying the TREM2 p.Q33X mutation display dysregulation of lysosomal function, reduced lipid droplets, and downregulation of cholesterol genes. The defective activation and downregulation of HLA-DR molecules together with the reduction in the lipid droplets protein PLIN2 detected in NHD iMGLs were rescued by enhancing lysosomal biogenesis through mTOR-dependent and independent pathways. Alteration in the same pathways were also observed in post-mortem brain tissues from NHD patients, closely recapitulating in vivo the phenotype observed in iMGLs in vitro.
Thus far, murine models are not able to recapitulate the key phenotypic features observed in patients affected by NHD. This has become an increasingly pressing challenge as rare variants in
TREM2 have been identified as risk factors in Alzheimer’s disease and frontotemporal dementia [
9,
37,
38,
42,
54]. Stem cell technologies have allowed us to obtain human cells from clinically defined patients, allowing for in-depth phenotypic analyses in exceedingly rare diseases such as NHD. Therefore, leveraging iMGLs represents a unique modeling platform to define disease-relevant, patient-specific phenotypes in a dish to understand disease mechanisms and for eventual therapeutic target identification, validation and drug discovery pipelines.
Transcriptional and functional analyses highlight that NHD iMGLs in vitro are hyporeactive compared to iMGLs from healthy controls, showing a deficit in HLA-DR, CD80, and CD86 molecules expression. This phenotype is reminiscent of what has been described in iMGL-chimeric mouse models harboring
TREM2 deficient microglia [
67]. Single cell RNAseq on iMGLs isolated from engrafted mice showed that
TREM2 KO iMGLs were shifted toward a more homeostatic profile even in wild-type mice, with a specific decrease of genes involved in the MHC-II and HLA presentation. Thus, the absence of TREM2 maintains iMGLs in a quiescent state, but this dormant state is released after LPS activation [
67]. Conversely, in NHD and controls brain tissues, NanoString analysis revealed the upregulation of genes involved in immune cell recruitment and activation, inflammation, microgliosis and astrocytosis. Analysis of overlapping genes between the microglial cluster from snRNA-seq dataset in Zhou Y at el. [
108] and our NanoString dataset revealed the downregulation of genes involved in lysosomal function, lipid metabolism and microglia activation. Upregulation of genes involved in inflammation and activation captured by NanoString analysis may be due to other brain cells (i.e. astrocytes, perivascular cells). The discrepancy between the inflammatory status observed in NHD tissues and the hyporeactive phenotype detected in iMGLs in culture, may be explained by the nature of the model system, where iMGLs capture function and transcriptome of cells isolated in culture, while post-mortem tissues are an end-stage snapshot resulting from the interaction of microglia with a much more complex cellular milieu after years of accumulating pathology. One possible scenario is that NHD microglia could be hypoactive early during the disease, and this abortive activation in NHD brains could lead later to a dystrophic and pro-inflammatory microglial phenotype. The hyperactivation observed in NHD iMGLs in vitro after LPS challenge indeed supports this possible scenario. To further support this hypothesis, in a model of acute CNS demyelination, lack of TREM2 was associated with defective microglia activation, while in chronic demyelination, TREM2 deficient microglia acquired a more pro-inflammatory and potentially neurotoxic phenotype [
13].
Multiple groups have examined the impact of NHD mutations on cellular phenotypes in iPSC-iMGLs models. These efforts have focused on the NHD mutations
TREM2 p.T66M, p.W60C, p.Y38C, p.V126G [
12,
18,
32,
50,
53,
73,
80]. Despite differences in the protocols adopted to generate human microglia, a common phenotype reported includes a reduction in TREM2 and sTREM2 levels, defects in phagocytosis of apoptotic bodies, and reduced survival in the mutant microglia. LPS-mediated cytokine secretion was comparable between control and
TREM2 missense mutations in one study [
32], while another group came to the opposite conclusion [
12]. Microglia-like cells from patients carrying
TREM2 p.T66M
, in either homozygous and heterozygous state, or
TREM2 p.W50C mutations, responded to LPS stimulation and were phagocytically competent despite accumulation of an immature form of TREM2 that was not trafficked to the plasma membrane [
12]. We observed common gene signatures shared by
TREM2 p.Q33X carriers and
TREM2 deficient iMGLs [
67]. These findings suggest that some, but not all, of the observed defects are driven by
TREM2 loss of function. Our sequencing results and functional experiments also highlight the differences between NHD1 and NHD2 iMGLs. This may be due to differences in the clinical severity of disease, other genetic modifiers, and/or sex-based microglial effects. Also, NHD1 and NHD2 iMGLs cultures exhibited some cell death, which was more pronounced in NHD2 compared to NHD1 iMGLs. Even though this finding is consistent with prior literature, supporting the role of the TREM2 receptor in microglial survival, we acknowledge that cell death in NHD iMGLs could constitute a confounding factor. Further studies are required to address these observations. Thus, studies of
TREM2 mutations are essential for capturing the full complexity of disease.
We show that lysosomal dysfunction is a common phenotype shared across NHD iMGLs and NHD brains. NHD iMGLs exhibit reduced labelling of acidic organelles and proteolytic capacity of degradative vesicles along with an accumulation of undegraded material in multivesicular bodies. To this end, we hypothesize that the accumulation of unprocessed material within the MVBs observed in NHD iMGLs is a consequence of the lysosomal defects. Similarly, we observed lysosomal defects in NHD brains, as shown by a decrease in CD68 protein staining in the frontal lobe of a NHD patient and decreased transcription of genes implicated in lysosomal acidification (
ATP6AP2) and chaperone mediated autophagy (
LAMP2). LAMP1 protein expression, which marks degradative vesicles, was increased in the basal ganglia, implicating regional-specific differences in NHD pathology. The dysregulation of lysosomal genes and accumulation of lysosomal debris have been widely reported in lysosomal storage diseases [
87]. Lysosomal dysfunction has broad consequences on cellular function and may affect core signaling events of the cell [
5,
8]. In this regard, MHC class II molecules and recycling of peptides displayed by MHC class II molecules are assembled in the endo-lysosomal compartment [
17,
30]. Thus, the reduction observed in MHC I and MHC II genes and proteins in NHD iMGLs might be explained by the deficient endo-lysosomal transport preventing the correct assembly and delivery of these molecules at the membrane. By targeting lysosomal and autophagic processes through TFEB activation in an mTOR-dependent or independent manner, we fully restored HLA-DR levels in NHD iMGLs. TFEB colocalizes with master growth regulator mTOR complex 1 (mTORC1) on the lysosomal membrane [
94], and pharmacological inhibition of mTORC1, activates TFEB by promoting its nuclear translocation. TREM2 activates mTOR through DAP12 and/or DAP10, via recruitment of upstream mTOR activators such as PI3K and AKT. TREM2 sustains cell metabolism through rapamycin complex 1 and 2 (mTORC1 and mTORC2, respectively) signaling and defective mTOR activation in TREM2-deficient microglia is associated with a compensatory increase of autophagy [
101]. This is the first evidence that the lysosomal compartment could be a critical hotspot impacted during the disease by the presence of TREM2 mutations in human microglia, largely responsible for NHD pathophysiology. These findings suggest that the endolysosomal degradative pathway could be a novel and attractive therapeutic target for this disease or others characterized by altered TREM2 function in microglia [
26,
87].
TREM2 p.Q33X iMGLs show downregulation of lipid metabolism at the gene network and protein level. We observe a reduced amount of lipid droplets in NHD iMGLs compared to controls, both in the presence or absence of myelin. Lipid droplets are dynamic subcellular organelles required for storage of neutral lipids such as glycero-lipids and cholesteryl esters, and have been recently described as essential to metabolically support immune responses, antigen cross-presentation, interferon (IFN) responses, and production of inflammatory mediators [
74]. Alteration in myeloid cell activation status is connected to profound changes in lipid droplets numbers and composition. Lipid droplets have also been shown to accumulate in microglia during aging, being part of a characteristic microglial transcriptional signature called lipid-droplets-accumulating microglia (LDAM), which show defects in phagocytosis, release elevated levels of proinflammatory cytokines and reactive oxygen species [
64]. Observations of reduced lipid droplet content in NHD iMGLs are consistent with work showing that, upon demyelinating injury,
Trem2-deficient mice are unable to elicit the adaptive response to excess cholesterol exposure, form fewer lipid droplets than wild-type mice, and develop endoplasmic reticulum stress [
34]. This supports the hypothesis that the TREM2 receptor is required for lipid droplets biogenesis and to buffer cholesterol-mediated toxicity in mouse and human microglia. Another possible interpretation of our findings, is that lysosomal dysfunction in NHD iMGLs would lead to an altered degradation of these lipidic structures [
106]. Further, our finding that lipid droplets content in NHD iMGLs was rescued via mTOR-dependent pathways is consistent with prior evidence that mTORC1 inhibition stimulates lysosomal hydrolysis of phospholipids and results in an adaptive shift in the use of constituent fatty acids which promotes their storage in newly formed lipid droplets for energy production [
40].
Finally, we observed the downregulation of genes involved in neuroplasticity, synaptic function and microglia-to-neuron cross-talk in NHD brains versus controls. Microglial survival is controlled by IL34 and colony-stimulating factor 1 (CSF1), through CSF1R signaling activation [
23,
43]. Microglia, in turn, are critical for the regulation of neuronal activity and firing in a region-specific and microglia number-dependent manner [
4]. This is highlighted by observations that NHD patients do not only display defects at the microglial level but instead present with neuronal phenotypes and dysfunction that might be microglial-dependent and/or independent. This area of study raises many new questions about the impact of
TREM2 mutations on neuronal function during neurodegeneration and highlights how studying NHD might help us also understand microglia-to-neuron cross-talk and the impact of
TREM2 mutations on this interaction. Overall, our results suggest that NHD might be considered a lysosomal-storage disease with white matter alterations as first proposed in reports based solely on NHD pathology [
49] when the genetic cause of the disease was not known. Our study provides the first cellular and molecular evidence that lack of TREM2 in microglia leads to a defect in lysosomal function. A better understanding of how microglial lipid metabolism and lysosomal machinery are altered in NHD may provide new insights into mechanisms underlying NHD pathogenesis.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.